|
Taking Grid to Market: IBM Announces Ten Commercial Grid Offerings
By Charles King
IBM has made a trio of announcements concerning its grid
computing initiatives, including the introduction of nine grid computing
offerings targeting the aerospace, automotive, financial markets, government,
and life science industries. These offerings include:
▫
Financial
Markets: An Analytics Acceleration Grid to accelerate trading operations and
increase computational throughput, and an IT Optimization Grid to better
utilize computational and storage resources.
▫
Life
Sciences: An Analytics Acceleration Grid to increase the number of
calculations processed in drug discovery research, and an Information
Accessibility Grid to provide unified data access across existing
computational resources and assets.
▫
Automotive
and Aerospace: An Engineering Design Grid to help manage costs by optimizing
use of existing infrastructure, and a Design Collaboration Grid to enable
data sharing and distributed work flow across partners.
▫
Governments:
An Information Access Grid to help maximize resources used in data mining
applications, and to simplify data sources and access through a unified data
and file interface.
IBM is also conducting Grid Information Workshops to
help companies determine how grid solutions can affect their organizations.
Additionally, IBM announced that it had completed a
grid R&D project with Charles Schwab that the company claims reduced the
processing time on a financial application from more than four minutes to
fifteen seconds. IBM also announced that it has established master
relationships with grid middleware vendors Platform Computing and DataSynapse,
as well as working relationships with Avaki, Entropia, and United Devices.
Net/Net
Grid computing may qualify as one of the most highly
trumpeted and least understood technologies around. In theory, grid solutions
can be used to pool all of a company’s heterogeneous IT resources, from the
desktop to the data center, allowing them to be viewed, managed, and
allocated as a single computing entity. Not only can this help improve
overall IT efficiency and cost effectiveness, but it can also provide
enterprises a means of performing complex computational tasks by distributing
them across existing IT infrastructures rather than on costly specialized
hardware.
Understandably, this notion has lately caused
something of a buzz in the industry, with major vendors including IBM, HP,
Microsoft, Cisco, Oracle, and Sun queuing up to declare fealty to various
grid-related standards efforts such as Globus, Legion, and the Global Grid
Forum, and promising to play nice together in order to deliver on grid’s
utopian promise. Along with major vendor efforts, grid has inspired more than
the usual share of tectonic industry rumbles and even been christened the
“next Internet” by a few IT soothsayers.
Is there any substance to all this high tech
styling? In truth, yes. While most of the buzz surrounding grid is relatively
new, computational grids have been around for awhile, both in voluntary
models (such as the SETI @Home project) and in solutions designed and
deployed by the grid middleware specialists mentioned in IBM’s announcement.
In fact, the list of enterprises that have deployed grids developed by
Platform Computing, United Devices, Entropia, and others include mainline
enterprises such as Monsanto, IncyteGenomics, Pacific Life Insurance,
Novartis Pharmaceuticals, and Gateway Computer.
So if grid appears to be for real, where does IBM’s
announcement fit into the greater scheme of things? First, it should be
pointed out that that by their nature, grid solutions need to be measured by
what is both possible and practical. Sure grids can be built and many have
been and are being successfully used across a range of IT-enabled industries.
But despite the hype/enthusiasm, grid computing remains for the most part a
highly complex practice whose solutions require an enormous amount of
customization. Additionally, while major IT vendors have made a good deal of
strategic noise about grid, few have delivered differentiated grid solutions.
HP, for example, has declared grid to be a major element of the company’s
still-evolving Utility Data Center solution. Sun has been quite aggressive
about the usability of its Grid Engine solution, but the vast majority of its
deployments have been comparatively small.
What makes IBM’s announcement particularly
interesting is the company’s unique focus on creating commercial grid
solutions to fit specific business/technical challenges. Does this mean that
IBM’s targeted grid offerings will be a slam-dunk success? Not at all. We see
two potential dangers lurking here. First, the practical and financial benefits
of grid are difficult to quantify, making it a particularly hard sell during
tough economic times. Additionally, since any new market bears a fair share
of buyer skepticism, grid’s inherent complexity could work against it. This
does not mean that making a market for grid will be impossible, just
difficult. However, by focusing its initial offerings to a few key grid
applications and specific industries, IBM is following a deliberate path to
highly directed product and market development that we believe could offer
the company significant future success.
|